In the Wake of Genocide: Designing a Rwandan Constitution

Rwanda faces the dual problems of ethnic division and economic underdevelopment. In early April 1994, a coterie of Hutu elites, faced with the invasion of the Rwandan Patriot Front (RPF) and fearing a dramatic shift in power, mobilized Rwanda’s eighty-percent Hutu majority to commit genocide against the twenty-percent Tutsi minority. The killings proceeded for three and half months, at a rate of more than twenty thousand a day – five times faster than the Nazi death camps. The RPF then took power, and more than two million Hutus fled to the Eastern Kivu region of the Congo, where violence continues to this day.

The Hutu-Tutsi ethnic divide at the center of the Rwandan Genocide is old but not ancient. The two ethnic groups, ascriptively and culturally similar – sharing physical semblance, language, heritage and religion – provide a textbook example of how ethnic division can be made salient by institutional design, and then made violent by political opportunism. From the 1930s to 1961, Belgian colonizers created institutions that specifically allocated political power along ethnic lines: using a card-based identification system, they stripped Hutus of their land, created a shadow extraction government headed by the Tutsi aristocracy, and gave them exclusive rights to tax collection power and state-funded education. Then in 1960, at the eve of Independence, the Belgians held an election in which they endorsed the Hutu politician Dominique Mbonyumutwa, affecting a precipitous change in political power that uprooted the monarchy headed by Tutsi aristocracy (Lec. 8.1 Levitsky). It is the major intention of this report to consider ways to reverse the very institutional incentive for ethnic identification that helped to create and perpetuate the myth of ethnicity in the first place.

The second problem Rwanda faces is that of underdevelopment and weak state strength. Rwanda is the most densely populated country in Africa with hundreds of thousands of refugees currently migrating back from neighboring Burundi to the inner borderlands of Rwanda. Development and ethnic reconciliation are taking place in the condition of scarcity, raising the stakes for every individual involved.

Though reconciliation is treated as primary (for no governance at all can take place if the Tutsi refuse its legitimacy or the Hutus enact reprisals) there is no strict hierarchy of concerns. Neither the problem of ethnic conflict nor the problem of underdevelopment can be conceived of in absence of the other – effective governance will require ethnic cooperation, and in turn, there is evidence that economic progress tempers ethnic division (Lipset 229). The problems are not culturally endogenous. Despite decades of literature on the cultural and social requisites for modernization, there is a growing consensus that development can occur anywhere with a proper mixture of effective policies and a strong state (Fukuyama 255). Ethnic division, likewise, is a product both of ancient “dispositions” and incentives; ethnic identity may indeed fulfill a primordial human need for a “terminal,” unfaltering community (Horowitz 490), yet it is just as true that ethnic division is activated by external factors, unlikely to become salient or violent unless under conditions of fear, scarcity or political advantage. “Ascriptive identity,” wrote Donald Horowitz, “is heavily contextual” (Horowitz 499).

What follows is: institutions are very important. While they cannot prevent the Hutu majority in a democratic Rwanda from enacting a political recrimination on the Tutsis, institutions can allocate political power in such a way as to disincentive such behavior. Democracy, properly designed, is an ingenious tool, because as Guiseppe Di Palma wrote, “it is open and open-ended, and because none of its players lose once and for all and on all arenas” (298).

The institutional designed put forth is based on the premise that the Hutu-Tutsi division was accentuated by institutional design at its outset, and thus, to be ameliorated, must be blurred rather than rigidified by constitutional institutions. Consociational systems such as those proposed by political scientist Arend Lijphart would fail Rwanda for exactly this reason. By institutionalizing ethnic division according to political power, consociational systems – ethnic-based coalition seats, federal autonomy and ethnic veto power – would further divide an already relatively homogenous society. Arend Lijphart explains that consociationalism functions best in “segmented pluralist” (498) societies, those with “fragmented culture” and “little or no overlapping between its distinct subcultures” (501). Each group in such a society may live side-by-side, yet totally autonomously, each with “its own religion, its own culture and language, its own ideas and ways” (504). Rwanda is not, by this standard, a plural society. Whereas the divisions in Lebanon (where consociational democracy was successful employed from 1947 until 1973) between Maronite Christians and Sunni, Shia Muslims are grounded in ascriptive physical properties, history, language and theology, the Tutsi and Hutu populations in Rwanda are phenotypically similar (vice president of the current National Assembly Laurent Nkongoli has famously said: “You can’t tell us apart; we can’t tell us apart”), linguistically similar, culturally and historically united and geographically inter-dispersed.

Thus, the following constitutional arrangements pursue a strategy of ethnic accommodation based on blurring – rather than accentuating – ethnic divisions between Hutus and Tutsis. In addition, they place high priority on the creation of an effective governing body. To achieve these ends, five principles will be observed*:

- First, promote fluid, ad hoc legislative coalitions over a rigid, majoritarian party system. The Hutu majority will be fragmented and the Tutsi minority will be granted full representation through a proportional representation (PR) voting scheme.

- Second, promote moderation between the Hutus and Tutsis by creating real political incentives for cross-ethnic appeal. A semi-presidential system will be enacted that demands, on the one hand, wide, centrist appeal for the executive branch (which will be voted in by a preferential voting system), and, on the other hand, cross-ethnic coalition building within the legislature. In addition, a portion of the legislature will be made up of “district representatives” voted in by plurality at an election concurrent to that of the executive, promoting a wide-appeal coalition for these members, similar to the coalition that propelled the executive. Finally, a ten percent quota would be placed on district PR elections to prevent extremists from gaining seats in the legislature.

- Third, provide for fair representation of both Hutu and Tutsi by maintaining a low disparity between votes caste and seats awarded and by creating a number of dispersed points of power, preventing a zero-sum game or a majoritarian suppression. This will be achieved through mixed-member PR (plurality and PR) and semi-presidential arrangements.

- Fourth, create standards of accountability. The president as well as the district representatives will be directly accountable to constituents in a way that a pure Parliamentary system or a pure PR system would not allow.

- Fifth, promote legislative effectiveness, achieved through the important mixture of long-term policy decisions and broad coalition of the executive with the fluid coalitions of the legislature, directed by the prime minister.

These arrangements will interact in productive ways, the whole greater than the sum of its parts. In its entirety, the constitutional arrangement will involve two key institutions:

1. Semi-presidential government where the executive is elected with a preferential voting system.
2. Mixed-member proportional representation, where twenty “district representatives” are voted in by district according to plurality in an election that occurs concurrent with the presidential election, and one hundred representatives are voted in according to a proportional representative system with a ten percent quota for entrance.

Semi-Presidential

First, a semi-presidential system will be implemented, where a popularly elected president (according to a preferential vote) will be, in Giovanni Sarotori’s phraseology, a “first among unequals” (446), sharing power with a prime minister. The prime minister is selected by the legislature and may be removed with a vote of no confidence. The president cannot be removed by the legislature with a mere no confidence vote (though impeachment, under certain circumstances will be permitted), and at the same time, cannot remove the prime minister. Neither branch has veto power over the other; thus there is a sharing rather than a separation of powers: both the president, who was elected by popular vote, and the prime minister, who is in his position by legislative coalition vote, share the daily governance of the county.

Pure parliamentary lacks long-term policy guidance and accountability; and pure presidential systems are crippled under the necessary (for Rwanda) conditions of multipartism. The great benefit of the semi-presidential system is that it provides for and protects against both: there is a strong executive that is broadly supported and accountable to the people, while at the same time functions well with the very fluid, divided coalitions of the legislature.

The importance for Rwanda of the semi-presidential system is grounded in the nature of the coalitions that the system promotes. Two separate coalitions sustain the president and the prime minister: the latter created after elections, according to logrolling that transpires in the halls of the parliament, while the former is created prior to elections, according to the popular vote of the people. This allows, as I will propose below, for the president’s moderate coalition to exert a directional influence on the fluid, ad hoc coalitions within the legislature, while never exerting total control over the government. Interests will co-mingle in fluid and diverse ways and accountability will be dispersed. The results will be: 1) that no one group gains a permanent majority, because the president’s support is compositionally different from the prime minister’s at any given moment, 2) compromise is promoted between two compositionally different coalitions, 3) promotes intra-ethnic competition, 4) creates a multiplicity of power sources, in both of the executive and the legislative, lowering the stakes and reducing the number of losers and finally 5) creates a strong incentive for the prime minister to create effective legislative coalitions (for he is directly accountable to the legislature for his continued tenure) and for the president to oversee good governance (because he directly accountable to the people). As a whole, then, the system maximizes coalitional cross-pollination, fluidity, representation and dispersion.

But the major concern for Rwanda remains: how to achieve a multi-ethnic coalition at all? With an eighty percent Hutu majority, what stops the president from deriving all his support from the Hutus, at the expense and potentially demise of the Tutsi minority? How moderate and cross-ethnic the winning coalition is will depend on the extent that 1) Tutsis vote as a bloc, and 2) the Hutu party splits. There is strong historical evidence that the Tutsis, having always been either in power or persecuted, will vote as a coherent whole. On the other side, there are a number of reasons to believe that the Hutu party will split. For one, the presidency is too powerful a position for there not to be Hutu internal competition for the prize. In addition, the president’s platform is by definition broad enough to cut across multiple ideological and cultural spectrums. Hutus have different financial statuses, different regional loyalties and different opinions about the nature of economic development, international relations, foreign investment and so on. The genocide itself was caused by a dissociation like this, between moderates in congress who were engaged in peace treaties with the RPF and the radical clan that was interested in holding onto power. Thus there are historical as well as structural aspects of the presidency that indicate that the Hutus will split. Finally, in accordance with Duverget’s observation about plurality congressional district and countless scholars’ statements about presidential systems, competition for the presidency will tend to promote the concentration of power into two strong parties (for example, Lijphart 485); thus the fragmentation, if it happens at all, will likely be in a two-way split.

If the above is valid, then it is likely that three groups will be important for any given presidential election: Hutu group A and Hutu group B, each of which accounts for roughly forty percent of the population, and Tutsi voting bloc C. Tutsi group C, representing twenty percent of the population, is a considerable electoral prize to win, and moderation would seem to be the answer – but only so long as Tutsis decide to vote, and if they do, are willing to vote Hutu. If they do not vote or “throw their vote away” on a minority Tutsi candidate, then they withdraw from political relevance and the moderating effect is gone. One way to resolve this is to utilize a preferential voting system, such as the one implemented (with success) in Sri Lanka. In this system, each person ranks his top-choice candidates according to preference and if no candidate in the system receives a fifty-percent-plus-one majority of “number one” preference votes in the first round of ballot counting, then the candidate with the fewest votes in the pool is removed and all his ballots are distributed among the remaining candidates according to preference. This continues until one candidate reaches a majority. Such a system is promising: by allowing everyone to vote by personal preference, Tutsis will have a reason to go to the polls and thus be kept psychologically and political integrated. More importantly, it assures that all the Tutsi votes will end up falling onto one of the two Hutu parties. The smart party, therefore, is highly vested in winning the Tutsi’s “number two” preference. The more the Tutsi’s remain as a voting bloc, the more vital efforts to win their “number two” preference become, and the more moderation and reconciliation will result. The cost of extremism would simply be too great.

For these reasons, the president’s coalition will likely be moderate rather than radicalized. It will also necessarily be broad, because it needs to account for fifty percent of the population.

The semi-presidential system promotes another type of moderation: it generates multiple centers of power, thus reducing the importance of any one place in the system. The president will owe proportionality to those Tutsis in the coalitions that helped him get elected and likewise the prime minister will owe favors to the Tutsi legislators that hold sway over his coalition building. In contrast to Juan Linz’s argument that the president, feeling that he “posses independent authority and a popular mandate” is likely to be “imbued” with “a sense of power and mission” (412), the semi-presidential configuration ensures that the president share his power with the prime minister, reducing the singular control he has over the direction of the country. On the other hand, a pure parliamentary system, Mainwaring argues, has the potentially of giving “a disciplined majority” complete control over both the executive and legislative branch. “Here, more than in any presidential system, the winner takes all” (424). By contrast, the negotiations needed between the president and the prime minister ensure that no one person has unilateral control over the direction of the country, and many people have sway.

Finally, the system provides for more effective governance within the context of fragmented parties. Linz argues that American-style presidentialism creates a “dual legitimacy” between president and legislature, where “no democratic principle exists to resolve disputes between the executive and the legislature about which of the two actually resents the will of the people” (415). Under conditions of multipartism, this leads to impasse or, more ominously, to the desire of the executive to override the parliament, as in the case of Yeltsin in Russia or Fujimori in Peru. A semi-presidential system, by contrast, works optimally under circumstances of multipartism. The presidential coalition guides the direction of the country and the prime ministers coalition provides the specific legislative majorities necessary. These divisions are precisely the benefit of semi-presidentialism, according to Giovanni Sarotori: they “enfeeble the president and force him into cohabitation with a prime minister of a different party; but this engenders a strengthened premier, who can and will fill a coalitional majority for his government. Thus semi-presidentialism can solve the problem [of divided majorities] that pure Presidentialism cannot solve” (448). In the context of a fractured Hutu majority – as outlined above for the presidency and applied to the legislature below – semi-presidential system is the best to promote accountability, moderation, and good governance.

Mixed-Member Proportional Representation

Second, the composition of the legislature will be determined through a mixed-member proportional representation system, where one hundred “national legislators” are voted in through proportional representation in the five already-established Rwandan provinces with a ten-percent quota for entrance. An additional twenty “district representatives” will be elected according to a plurality system vote held concurrently with the presidential election.

The fact that Tutsis and Hutus are dispersed evenly through Rwanda means that under a plurality system it is likely that no Tutsis would have a seat in the government. PR is practically a necessity for such a country, with a twenty percent persecuted ethnic minority widely distributed. “PR was designed to provide minority representation,” writes Arend Lijphart. He quotes Stein Rokkan as saying that “It was no accident that the earliest moves towards proportional representation (PR) came in the ethnically most heterogeneous countries” (486). Statistics show that under PR this representational fairness – the low disparity between votes and governmental representation – is much higher than under plurality systems. “There is little doubt that this is indeed the case. For instance, where ethnic minorities have formed ethnic political parties, as in Belgium and Finland, PR has enabled them to gain virtually perfect proportional representation” (487). So important is PR for ethnic minorities that even the British government, with a long, vaunted history of Westminster-style majoritarianism recognized the need to integrate the Northern Irish population into the political system through PR (Lijphart 488).

Not only is PR important for representational fairness, but also absolutely essential for fragmenting the Hutu majority. As discussed above, the Hutu majority must be split in order for the Tutsis to be integrated into the political system. The more ways this split can occur within the legislature, the less likelihood that there will be any one majoritarian coalition that could exclude the Tutsis. For example, a two-way split in the Hutu, while likely for the election of the president, would be dangerous if it existed in the legislature, for the two majority parties might compete for power on most issues, but agree to align against the Tutsis on ethnically sensitive ones. The more splits in the system, the more numerous and fluid are the coalitions, and the more difficult it is to align against the Tutsis at any one vote; and, at the same time, the smaller the coalitions, the more valuable the twenty-percent Tutsi voting bloc.

Fragmentation of parties occurs in direct proportionality to the magnitude of district. “The number of parties winning seats, as well as the proportionality of electoral systems, generally rises along that continuum,” states John Carey (472). The more legislators in a district, the lower the percent-based “barrier of entrance” into the legislature, and thus the higher the incentives for minority parties to formulate and run. One nasty byproduct, however, of this low barrier of entrance is that extreminist parties might be legitimized by gaining seat in the legislature. “Without PR,” notes Guy Lardeyret, “the Communists and the Nazis would probably not have been able to storm onto the German political scene as they did in the 1930s” (492). It is vital that the Hutu majority fragment, and best if this is along multiple lines; yet extremists, such as those that helped to perpetrate the massacre, should not be given a vote. To negotiate this compromise, a ten percent quota will be placed onto entrance. Ten percent is sufficiently low for full Tutsi representation and for substantial Hutu fragmentation, but sufficiently high to prevent chaos and extremism.

An additional problem with PR is that it tends to reduce the accountability of the representatives – shifting their concerns from the region to the party. “The critical distinction,” writes John Carey, “is between institutional arrangements that encourage legislators to cultivate personal reputations among voters and those that put the collective reputations of the party first” (446). Carey argues that personalistic politics is bad, because it tends to promote pork-barrel legislation and legislators without broad sophistication on national issues. These are real concerns, but less important than the benefits of having some representatives that are directly accountable to their district. The election of twenty district representatives would make sure that local issues that concern Tutsi and Hutu alike are targeted – local infrastructure, school systems, irrigation and so on. Two other reasons for accountability: one that in an ethnically polarized state accountability reduces the chances of ethnic scape-goating; and two, it allows for democracy that takes place, as Lardeyret points out, “at the level closest to the citizen” (494).

Combination of the two

The final phase of the argument for this arrangement concerns the way that the two systems interact. The mix of both one hundred national legislators voted by PR and twenty district representatives voted by plurality would reflect the mix of the president with the parliament. The plurality district representatives would tend towards a two-way Hutu split, with moderate legislators with broad-majority appeal; the PR for the one hundred national legislators would create fluid coalitions that never rigidify along ethnic lines. Likewise, the president’s coalition would be broad and moderate, negotiated prior to election; and the legislature’s coalition would be ad hoc, negotiated after election, in the halls of congress. This analogy is deliberate, and would be reflected in voting arrangements. Carey writes that, if executive and legislature are voted in concurrently, “the effects of institutional rules governing executive elections tends to spill over into legislative party systems” (473). The district representatives (plurality scheme) will be voted in concurrently with the president so that the broad, multi-ethnic coalition of the executive influence the creation of moderate, multi-ethnic coalitions within the districts. The national legislators, on the other hand, will be voted at a date separate from the executive and district representatives, helping their elections to tend towards fragmentation.

No literature [that we have studied] has explored the possibility of aligning a mixed semi-presidential system with a mixed member proportional representation system, where the plurality reps are voted concurrently with the president. The effect of such an arrangement would potentially be to create a “shadow coalition” superimposed on the fragmented legislature. The broad, multi-ethnic group negotiated by the president on a national scale, and within every district by the district representatives could provide a long term multi-ethnic framework to guide the actions of the multi-ethnic coalitions created by convenience in the legislature. Figure 1 demonstrates a conceptual framework to this. At the very least, this system confuses the lines of allegiance and disperses power points, making it difficult for the Hutus to solidify against the Tutsis. At the best, there is at once fragmentation and unity; there is long-term direction and short-term flexibility.

The above institutions reflect the national need for the gradual dissolution of ethnic divisions. There is no guarantee that the institutions outlined above will actually result in what is argued: while I conjecture that two separate groups of coalitions would be created for electing the president (broad) and for electing the PR national legislators (fragmented), Rwanda could instead create a rigid two-party system, from which it would be difficult to break. Horowitz demonstrates through Guyana’s unsuccessful attempts to maximize both representation and fragmentation, that PR cannot split a “result already entrenched in the party system” (558). On the other hand, though presidential seats tend to concentrate power, the preferential voting scheme (still novel) may incentivize the fragmentation of power because everyone gets a vote, making the president’s coalitions less broad, perhaps in unexpected ways.

The system does, however, create incentives for cross-cutting allegiances, multiple points of power, and fluidity, which in turn promotes fair representation and moderation. Moreover, the constitutional arrangements provide for a sharing of power between president and prime minister that is optimal for effective government under circumstances of multipartism. Indeed, as it is hoped with the forgoing, Rwanda may find that the very types of constitutional institutions which can be pointed to as the source of the Hutu-Tutsi division may someday be pointed to for its solution.



* These are adapted in part from Donald Horowitz’s recommendations in Ethnic Groups in Conflict (1985), course book pages 542 and 549.